As it happens, I tried a 2 x 120V to 2 x 9V 50VA transformer as an audio transformer just the other day. If you do the sums, you find that the voltage ratio is appropriate (primaries in series, secondaries in parallel) for treating the transformer as a 5ka-a to 7R load. Note that SE operation is not permissible (unless you use an anode loading choke and capacitor couple to the output transformer). In a push-pull circuit, there is theoretically no DC magnetising current, and an ungapped core (toroid) is feasible.
I drove the transformer from the output of an oscillator from a source resistance of 2k4 (representative of typical triode conditions) and loaded the secondary with a 6R8 resistor. I applied a 10kHz square wave and achieved a surprisingly good result - significantly better than that achieved by a vintage EI Woden 20W output transformer under the same conditions. Even more surprisingly, the low frequency performance was good too, although I would expect the core to be unable to sustain 50Hz at >20W without significant distortion.
Before you get too exited, bear in mind that it will be very difficult to to persuade a practical push-pull output stage to balance the output stage currents to the degree needed by an ungapped core. This is the weakness. You save money on the output transformer, but pay for it in the bias arrangements....
I drove the transformer from the output of an oscillator from a source resistance of 2k4 (representative of typical triode conditions) and loaded the secondary with a 6R8 resistor. I applied a 10kHz square wave and achieved a surprisingly good result - significantly better than that achieved by a vintage EI Woden 20W output transformer under the same conditions. Even more surprisingly, the low frequency performance was good too, although I would expect the core to be unable to sustain 50Hz at >20W without significant distortion.
Before you get too exited, bear in mind that it will be very difficult to to persuade a practical push-pull output stage to balance the output stage currents to the degree needed by an ungapped core. This is the weakness. You save money on the output transformer, but pay for it in the bias arrangements....
commercial power toroids
One caution I have found about using commercial power toroid transformers. The 130 VA Avel-Lindberg Ltd xfmr I found surplus has the two 120 V primaries wound as random bifilar, ie, both windings were wound at the same time to save fixture setup times. This causes the two primary windings to have 21000pF capacitance between them. Also, there is little effective isolation from primary to secondary, with 850pf common mode. It measures impressive bandwidth from a 50 Ohm source, but not so hot with a realistic source impedance. So, I would check for this problem before using one in an amplifier.
One way to avoid this capacitance problem might be to use a 120V to 120V (or 240V,240V if you can find one) toroidal isolation xfmr, which would presumably have some insulation between the windings. Then wind your own 8 Ohm secondary on top. Unfortunately, most isolation transformers do not have the same number of turns on primary and secondary in order to compensate for voltage drop from winding resistance. So, may be hard to find a good xfmr, or will have to add some turns to one of the primaries to equalize.
Don
One caution I have found about using commercial power toroid transformers. The 130 VA Avel-Lindberg Ltd xfmr I found surplus has the two 120 V primaries wound as random bifilar, ie, both windings were wound at the same time to save fixture setup times. This causes the two primary windings to have 21000pF capacitance between them. Also, there is little effective isolation from primary to secondary, with 850pf common mode. It measures impressive bandwidth from a 50 Ohm source, but not so hot with a realistic source impedance. So, I would check for this problem before using one in an amplifier.
One way to avoid this capacitance problem might be to use a 120V to 120V (or 240V,240V if you can find one) toroidal isolation xfmr, which would presumably have some insulation between the windings. Then wind your own 8 Ohm secondary on top. Unfortunately, most isolation transformers do not have the same number of turns on primary and secondary in order to compensate for voltage drop from winding resistance. So, may be hard to find a good xfmr, or will have to add some turns to one of the primaries to equalize.
Don
Re: commercial power toroids
That's most surprising. I'd expect safety problems if that transformer was configured for 240V mains.
Yes, I've wondered about that, too. Although the Plitron output transformer specifications note that they were achieved with balanced DC currents, there is no mention of what imbalance (if any) might be tolerated.
Hacknet: For 7ka-a to 6R (a good choice for nominal 8R loudspeakers), that's an impedance ratio of 1167:1. The voltage ratio is the square root of the impedance ratio, so it's 34.1:1. If you have a 2 x 120V primary, that's 240V total, to you need 240V/34.1 = 7V. You won't find that, so I'd try a 6V transformer. If you can find a 2 x 110V with 2 x 6V 50VA transformer, then that should do nicely. Parallel the two 6V windings, but connect the 110V windings in series.
smoking-amp said:
That's most surprising. I'd expect safety problems if that transformer was configured for 240V mains.
SY said:
Yes, I've wondered about that, too. Although the Plitron output transformer specifications note that they were achieved with balanced DC currents, there is no mention of what imbalance (if any) might be tolerated.
Hacknet: For 7ka-a to 6R (a good choice for nominal 8R loudspeakers), that's an impedance ratio of 1167:1. The voltage ratio is the square root of the impedance ratio, so it's 34.1:1. If you have a 2 x 120V primary, that's 240V total, to you need 240V/34.1 = 7V. You won't find that, so I'd try a 6V transformer. If you can find a 2 x 110V with 2 x 6V 50VA transformer, then that should do nicely. Parallel the two 6V windings, but connect the 110V windings in series.
EC8010 said:
got a hack saw? gap the core ! as I pointed out on the National Semi website, Sanjay Maniktala has an excellent outline of the equations needed to work with gap sizes:
http://www.national.com/onlineseminar/2004/magnetics/Magnetics_033104.pdf
Re: Re: Re: commercial power toroids
Who knows? A toroidal main transformer will certainly be cheaper than an output transformer designed for the job. I suspect in the end that it will be down to whether you can keep those anode currents in balance. If you can, then the HF performance will probably be better.
hacknet said:
Who knows? A toroidal main transformer will certainly be cheaper than an output transformer designed for the job. I suspect in the end that it will be down to whether you can keep those anode currents in balance. If you can, then the HF performance will probably be better.
bifilar primaries
"That's most surprising. I'd expect safety problems "
Yes, I agree. I was surprised to find it was wound that way too. Probably why it was being sold surplus. It was coated in epoxy so was not obvious until I tore it down. I mentioned it to someone in the power supply industry and they thought it was a poor idea too. Since then, however, I have been even more shocked to find that the low cost E-I dual bobbin transformers sold by the likes of Signal, Stancor, Magnetek ... etc all have their dual 120V primaries wound bifilar on one of the bobbins. These are most likely all made in China. I also had a Jameco low cost E-I isolation xfmr which buzzed and had high common mode capacitance, so I tore it down to see how it was made, and it had restamped laminations full of holes and cutouts with no insulation between primary and secondary. Probably from India. Apparently old scrap laminations are shipped over there and they punch out smaller lams from the pieces. Transformer quality is getting scary.
For that reason I have been accumulating some large toroid cores from the local scrap yard with the idea of re-winding them myself for audio outputs (bigger core, fewer turns). The secret of getting large bandwidth from toroids is actually well known and published, and far simpler than the interleaving complexities of E-I xfmrs. Each winding must be wound using progressive or chevron winding technique. This is done by dithering the core back and forth slightly as the winding is done in a single overall pass around the core (on automated equipment). The winding piles up in short chevron shaped layers with the result that all turns are always near turns with little voltage difference. This minimizes distributed capacitance. For DIY, can just use a lot of multi-layered 1/2 inch wide mini-sections with an insulating sheet between adjacent section ends. Good insulation between the overall windings (layered) (and core too) is needed to minimize common mode capacitance. By making each chevroned or mini-sectioned winding cover 95+% of the core, leakage inductance is minimized. The toroidal core with no lamination gaps allows one to use higher maximum flux density as well (10,000 G versus 7500 G ,M6, for E-I) for higher power rating (with same linearity) and has higher effective permiability for higher primary shunt inductance. You can expect to get near 1/2 MHz bandwidth this way.
Don
"That's most surprising. I'd expect safety problems "
Yes, I agree. I was surprised to find it was wound that way too. Probably why it was being sold surplus. It was coated in epoxy so was not obvious until I tore it down. I mentioned it to someone in the power supply industry and they thought it was a poor idea too. Since then, however, I have been even more shocked to find that the low cost E-I dual bobbin transformers sold by the likes of Signal, Stancor, Magnetek ... etc all have their dual 120V primaries wound bifilar on one of the bobbins. These are most likely all made in China. I also had a Jameco low cost E-I isolation xfmr which buzzed and had high common mode capacitance, so I tore it down to see how it was made, and it had restamped laminations full of holes and cutouts with no insulation between primary and secondary. Probably from India. Apparently old scrap laminations are shipped over there and they punch out smaller lams from the pieces. Transformer quality is getting scary.
For that reason I have been accumulating some large toroid cores from the local scrap yard with the idea of re-winding them myself for audio outputs (bigger core, fewer turns). The secret of getting large bandwidth from toroids is actually well known and published, and far simpler than the interleaving complexities of E-I xfmrs. Each winding must be wound using progressive or chevron winding technique. This is done by dithering the core back and forth slightly as the winding is done in a single overall pass around the core (on automated equipment). The winding piles up in short chevron shaped layers with the result that all turns are always near turns with little voltage difference. This minimizes distributed capacitance. For DIY, can just use a lot of multi-layered 1/2 inch wide mini-sections with an insulating sheet between adjacent section ends. Good insulation between the overall windings (layered) (and core too) is needed to minimize common mode capacitance. By making each chevroned or mini-sectioned winding cover 95+% of the core, leakage inductance is minimized. The toroidal core with no lamination gaps allows one to use higher maximum flux density as well (10,000 G versus 7500 G ,M6, for E-I) for higher power rating (with same linearity) and has higher effective permiability for higher primary shunt inductance. You can expect to get near 1/2 MHz bandwidth this way.
Don
